Abstract
We study social learning in markets where a monopolistic seller and potential buyers learn from past buyers’ reviews and experiences. Because buyers can wait before purchasing, they have an incentive to free-ride on information from other buyers. The seller has to take this informational externality between buyers into account in her pricing. Our main result is that any Markov perfect equilibrium is inefficient because of too little and too slow sales. This is contrary to the results absent learning. We then show that the seller’s commitment power can alleviate these inefficiencies. In addition, we demonstrate that good news learning causes too little total sales and bad news learning causes too slow sales.
| Original language | English |
|---|---|
| Article number | 106136 |
| Peer-reviewed scientific journal | Journal of Economic Theory |
| Volume | 232 |
| ISSN | 0022-0531 |
| DOIs | |
| Publication status | Published - 01.01.2026 |
| MoE publication type | A1 Journal article - refereed |
Keywords
- Coase conjecture
- Durable goods
- Experimentation
- Exponential bandits
- Social learning